Hi Albrecht: On 02/09/2019 06:54:42 AM Sat, Albrecht Dreß wrote:
Am 08.02.19 18:52 schrieb(en) Albrecht Dreß:After digging through the code, I /think/ the problem is in libbalsa_mailbox_imap_fetch_structure() which loads only text/* and short messages completely (to be honest, I don't understand why multipart/* is working, though…). At least, the attached trivial patch, just loading single-part S/MIME messages just as text/*, solves the issue for me.Thinking again about this issue, I guess this approach is not completely correct. If all multipart/* messages are handled at some other place (are they?), shouldn't *all* other top-level content types be loaded here? The vast majority will (explicitly or implicitly) be text/*, and application/pkcs7-mime is already somewhat special. But a message containing only a, say image/* or application/pdf as the only payload is absolutely legal (although I saw the latter with malspam attacks only…). Thus, the more appropriate approach would be --8<-------------------------------------------------------------------------------------- diff --git a/libbalsa/mailbox_imap.c b/libbalsa/mailbox_imap.c index d85e377c1..2ba010b43 100644 --- a/libbalsa/mailbox_imap.c +++ b/libbalsa/mailbox_imap.c @@ -2247,7 +2247,7 @@ libbalsa_mailbox_imap_fetch_structure(LibBalsaMailbox *mailbox, LIBBALSA_MESSAGE_GET_LENGTH(message)<8192 || (message->headers && (!message->headers->content_type || - g_mime_content_type_is_type(message->headers->content_type, "text", "*"))) ){ + !g_mime_content_type_is_type(message->headers->content_type, "multipart", "*")))) { /* we could optimize this part a little bit: we do not need to * keep reopening the stream. */ GMimeStream *stream = --8<-------------------------------------------------------------------------------------- wouldn't it?
This version makes sense to me. Committed and pushed. We could always limit it to application/pkcs7-mime if there are issues… Best, Peter
Attachment:
pgpTTsfZnI4Cq.pgp
Description: PGP signature